Methods and systems for extended spatial scalability with picture-level adaptation

1. A method for picture-level adaptation in a spatially-scalable video codec, said method comprising: a) receiving a definition of a first cropping window at a first location in a first picture of an input video sequence; b) generating a first base layer corresponding to said first cropping window; c) receiving a definition of a second cropping window at a second location, not coincident to said first location, in a second picture of said input video sequence; and d) generating a second base layer corresponding to said second cropping window; and e) predicting an area of said second picture in said second cropping window by adjusting a motion vector to account for the difference between said first location of said first cropping window and said second location of said second cropping window.

2. A method as described in claim 1 further comprising predicting an enhancement layer for said second picture based on said second base layer.

3. A method as described in claim 2 wherein said predicting comprises use of a scaled motion vector having an image motion vector component and a cropping window motion vector component.

4. A method as described in claim 2 wherein said predicting comprises a motion vector mv=(d x , d y ) that is scaled in the vector mv s =(d sx , d sy ) using the following equations: { d sx = ( d x · w extract + sign ⁡ [ d x ] · w base / 2 ) / w base + 4 · ( x orig , r - x orig ) d sy = ( d y · h extract + sign ⁡ [ d y ] · h base / 2 ) / h base + 4 · ( y orig , r - y orig ) wherein w extract and h extract are the width and height, respectively, of a sub-region of an enhancement layer picture, w base and h base are the width and height, respectively, of a base layer picture corresponding to the sub-region of the enhancement layer picture, x orig and y orig are the coordinates of the origin of the sub-region of the enhancement layer picture, x orig,r and y orig,r are the coordinates of the origin of the entire enhancement layer picture and wherein sign[x] is equal to 1 when x is positive, (−1) when x is negative, and 0 when x equals.

5. A method as described in claim 2 wherein said first picture and said second picture are texture pictures.

6. A method as described in claim 2 wherein said first picture and said second picture are residual pictures.

7. A method for predicting an enhancement layer area, defined by a user-defined, cropping window, from a base layer in a spatially-scalable video codec, said method comprising: a) receiving a first location for a user-defined, cropping window in said enhancement layer of a frame of said spatially-scalable video codec; b) calculating a base-layer motion vector in a direction for an image block in said base layer; c) scaling said base-layer motion vector based on the ratio of a dimension of said cropping window in said direction to a dimension of said base layer in said direction; and d) adjusting said scaled motion vector to account for said location of said cropping window relative to said base layer.

8. A method as described in claim 7 further comprising predicting said enhancement layer using said adjusted, scaled motion vector.

9. A method as described in claim 7 wherein said scaling and said adjusting are performed in two orthogonal directions.

10. A method as described in claim 7 wherein said scaling and said adjusting are performed on a motion vector mv=(d x , d y ) that is adjusted and scaled in the vector mv s =(d sx , d sy ) using the following equations: { d sx = ( d x · w extract + sign ⁡ [ d x ] · w base / 2 ) / w base + 4 · ( x orig , r - x orig ) d sy = ( d y · h extract + sign ⁡ [ d y ] · h base / 2 ) / h base + 4 · ( y orig , r - y orig ) wherein w extract and h extract are the width and height, respectively, of a sub-region of an enhancement layer picture, w base and h base are the width and height, respectively, of a base layer picture corresponding to the sub-region of the enhancement layer picture, x orig and y orig are the coordinates of the on in of the sub-region of the enhancement layer picture, x orig,r and y orig,r are the coordinates of the origin of the entire enhancement layer picture and wherein sign[x] is equal to 1 when x is positive, (−1) when x is negative, and 0 when x equals.

11. A method as described in claim 7 wherein said prediction comprises textural prediction.

12. A method as described in claim 7 wherein said prediction comprises residual prediction.

13. A method as described in claim 7 wherein said scaling and said adjusting are applied to a all motion vectors of a prediction macroblock MB_pred.

14. A system for predicting an enhancement layer area, defined by a user-defined cropping window, from a base layer in a spatially-scalable video codec, said system comprising: a) a computing device for calculating a base-layer motion vector in a direction for an image block in said base layer; b) said computing device further comprising a scaling function for scaling said base-layer motion vector based on the ratio of a dimension of said user-defined, image-content-independent cropping window in said direction to a dimension of said base layer in said direction; and c) said computing device further comprising an adjusting function for adjusting said scaled motion vector to account for motion of said user-defined, cropping window from a location of a second user-defined, cropping window in a reference picture from which said predicting is being performed.

15. A method as described in claim 14 further comprising a predictor for predicting an enhancement layer using said adjusted, scaled motion vector.

16. A method as described in claim 14 wherein said scaling and said adjusting are each performed in two orthogonal directions.

17. A method as described in claim 14 wherein said scaling and said adjusting are performed on a motion vector mv=(d x , d y ) that is adjusted and scaled in the vector mv s =(d sx , d sy ) using the following equations: { d sx = ( d x · w extract + sign ⁡ [ d x ] · w base / 2 ) w base + 4 · ( x orig , r - x orig ) d sy = ( d y · h extract + sign ⁡ [ d y ] · h base / 2 ) h base + 4 · ( y orig , r - y orig ) wherein w extract and h extract are the width and height, respectively, of a sub-region of an enhancement layer picture, w base and h base are the width and height, respectively, of a base layer picture corresponding to the sub-region of the enhancement layer picture, x orig and y orig are the coordinates of the origin of the sub-region of the enhancement layer picture, x orig,r and y orig,r are the coordinates of the origin of the entire enhancement layer picture and wherein sign[x] is equal to 1 when x is positive, (−1) when x is negative, and 0 when x equals.

18. A method as described in claim 14 wherein said prediction comprises textural prediction.

19. A method as described in claim 14 wherein said prediction comprises residual prediction.

20. A method as described in claim 14 wherein said scaling and said adjusting are applied to a all motion vectors of a prediction macroblock MB_pred.